blog.ebg13.at

I went to this talk today. As it lies within my natural habits of being unable to keep my mouth shut, I had to comment.

Dr. Zemaneck starting telling tales from the stays with the first computers. How they translated their programs to binary before loading it into the computer, how he loved the sound about the rattling of the Mailüfterl, how it clicked and clicked away at every instruction. He mentioned that back then, what you could do was very limited compared to what you can do today. He’s amazed about how he can look at maps from all over the world, look at videos. Yet, he can still hear the computer rattling.

Dr. Chroust continued with his story on the beginnings of software development, the first programming languages and so on and so on… basically the same story we heard in the Software Engineering I and Allgemeine Systemlehre. We’ll old man just loves him self for the wonderful things he did.

Dr. Mössenböck gave a very nice review of how programming languages have evolved over the years. His experiences with the various languages, Cobol and Fortran, how they developed a 7 pass compiler at first because they only had 128k of memory. I really respect Dr. Mössenböck because I know of the work he has done.

Mag. Rabiser started off telling us how he got to programming. He started with a computer in his mom’s office that started with a DOS shell uhuhuhuhuu.. fear! When they first got a computer, it was a used notebook, 286, 10mhz, something like that… monochrome screen. poor guy. they still put down 80.000 schilling back then. He then explained to started studied business-computer-science and that that would explain the Mag. title instead of the Dipl.-Ing. Title. I don’t know if he was bragging or if he was sad about it. I won’t comment.

Anyhow, I really got upset when he mentioned that the best programming language is the one that one is best at. Everybody was like “uh-ah”, “so true”. Bullshit. For every domain there is a should be a good programming language. Some programming languages are really good at low-level programming, like C, some are better for large project with a lot of interconnected componentes (c++/java), some are good at string processing (Perl), some are good for file system scripting (Bash-script), the list goes on and on. It’s as if you’re in a racing tournament where you do one race on every kind of race track. You won’t be using a forumlar1-car in a rally-race etc.

We are now at a point, where programming languages seem to be able to do everything we need and the rest is being solved with tools: Generating Code, filling a few lines here, adding some there. Code reuse is brought to a maximum. Yet I still see programs crashing, not scaling to multiple processors, having to be stopped to be upgraded, the list goes on (readers familiar with a modern functional programming language I won’t name right now). Why don’t we leave that old procedural paradigm behind and follow the path to the light?

I brought up the discussion about what’s the next step in programming. There are quite a few researchers really thinking about the problem programmers have with software stability, especially with multi-threaded applications. They analyzed the most common sources for errors in languages and tried to take them away from the language. This resulted in a programming called Erlang. While I haven’t really writting big projects with it, I’ve been reading Joe Armstrong’s Programming Erlang book and am truly convinced by the concepts of the language.

Then came the moment when I lost a lot of respect for Dr. Chroust: He somehow heard Erlang was a functional programming language and immediately had to comment on performance. “When I got my first laptop, it had 40k auf hard-diskspace!” “you have to think about space usage” bla bla bla. I just commented that at some point nobody tought we would be sending XML documents across the internet, yet we do. Heck, we have lot’s of performance we could use, we just have to make it easier to write our programs with parallelism in mind.

Well, he said that our thinking is sequential. I found that really strange as well. We want to model things they way they are in the real world most of the time and yet we think sequential? Thats not true. And if we do, we have to start thinking in parallel. The world is parallel!

Next he mentioned that they already found out that there is a maximal amount of parallelism you can acheive. Heck yes, everybody having waited all summer for a grade in his parallel computing course knows about Amdahl’s Law. And while Amdahl is basically right that in any given program you can only parallelize a specific amount of instructions, we have to realize that we can handle bigger problems by utilizing more processors. So his law is not wrong, it’s just not relevant. Somebody like Dr. Chroust should have realized this. I am really disappointed.

Side-note: This post is the opinon of Philipp Aumayr. None of the other ebg13 members are responsible for this.

This entry was posted on Friday, September 28th, 2007 at 20:08 and is filed under IT, rants. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.